Animal collection
We collected newly oviposited wood frog egg masses from nine roadside ponds in upstate New York, USA that varied in their chloride concentrations at the time of egg collection, ranging from 1 to 744 mg Cl-/L. Given that salinity effects tend to diminish beyond 50 m, all ponds were less than 40 m from the nearest road. During 6-12 April 2022, we gathered ten partial egg masses (i.e. approximately one-fourth of a mass) from eight sites and seven partial masses from one site where only seven masses were laid. We placed the partial egg masses in 350-L outdoor wading pools (one wading pool for each population) containing aged well water at the Rensselaer Aquatic Laboratory in Troy, NY. All populations began hatching between 14 and 15 April 2022. Once they hatched, we fed the tadpoles rabbit chow ad libitum (Blue Seal Fresh Show Hutch Deluxe). We used hatchling tadpoles from these egg masses in a time-to-death experiment followed by a growth and development experiment. At least 1 wk prior to each experiment, we brought tadpoles from each population into the lab to acclimate. In the lab, the animal-rearing room was maintained at 21°C with a light:dark cycle of 15:9 hrs.
Time-To-Death Experiment
To assess the relative levels of tolerance of each population to road salt contamination, we conducted a time-to-death (TTD) experiment where we exposed tadpoles to control water (19 mg Cl-/L) and a lethal concentration of 8 g/L of NaCl. Time-to-death experiments intentionally use high concentrations of a pollutant (often higher than experienced in nature) to determine whether different groups in individuals (e.g., populations) differ in their relative tolerance over a short time period (1–4 d), during which the animals are not fed. Once TTD experiments find differences in tolerance based on short-term times to death, researchers can examine environmentally relevant concentrations over longer periods of time to determine if differences in tolerance affect performance.
For this experiment, we used a fully randomized design that involved the nine populations of wood frogs and the two salt treatments. We placed one tadpole into each of the experimental units, which were 59-ml clear plastic cups filled with approximately 45 ml of water. The lethal salt dose was created by adding lab-grade NaCl to aged tap water to reach a concentration of 8 g/L NaCl. From each population, we placed 15 individuals into cups containing water with a lethal salt concentration and five individuals into cups containing control water. Thus, there was a total of 180 cups in a completely randomized design. The control water was tap water that we aged at least 24 hrs to allow any chlorine to off-gas. We also set aside ten individuals from each population to estimate the initial average mass and developmental stage of each population. The initial individual masses of the populations ranged from 33 to 64 mg; developmental stages ranged from approximately 26 to 27.
We started the experiment on 26 April and team members took turns initially checking the tadpoles every 4 hrs for mortality. Once mortality began, we checked the experiment every 2 hrs and recorded the number of dead tadpoles from each population. We checked for deaths by gently blowing water at the tadpoles using a pipette. Nonresponsive tadpoles were considered dead. If it was unclear whether a tadpole was still alive, we looked for a heartbeat under a dissecting microscope. We set aside individuals that were declared dead and double checked them at the subsequent check time. If we determined that the tadpoles were still alive, we returned them to the experimental array. Checks continued for a total of 96 hrs (30 April). There was no mortality in the salt -free controls. Upon completion of the experiment, we euthanized all individuals using 2% MS-222.
We used pairwise log-rank tests with a Bonferroni correction to look for differences in the Kaplan-Meier survival curves for each population. While this analysis tests for differences among the nine populations, it does not show what may be driving differences in survival. Thus, our analysis sought to determine the direct effect of chloride concentration of the source pond and the death rates of each individual. To do so, we conducted a Cox mixed effects model. Our response variable in this model was the probability of tadpole survival, regardless of the pond from which they came. Our fixed effects were the source-pond chloride concentrations, the mean individual mass of tadpoles in each population, and their interaction. We included population as a random effect in the model to control for other differences among populations, aside from chloride concentration, that we did not measure. We calculated hazard ratios for the fixed effects using the model parameter estimates. To do this, we raised the natural root to the power of the parameter values. Subtracting the hazard ratio from one gives the percent increase in survival that a parameter contributes to the overall survival of individuals in the study. Parameter significance was determined using Wald Chi-square analyses. All analyses were conducted in R version 4.1.0 using the survival, Survminer, coxme, and car packages.
Growth and Development Experiment
To determine whether evolved salt tolerance came with any growth or developmental tradeoffs, we conducted a follow-up experiment on additional wood frog tadpoles from the same nine populations. We used a randomized block design for this experiment with three spatial blocks, where each block was a different shelf height in the lab. In each block, experimental units were assigned one of the nine populations and one of two salt treatments (control water with no salt added vs. a higher sublethal salt concentration). We replicated the 18 treatment combinations twice within each block, which provided six replicates across the entire experiment. Thus, there were 36 experimental units per block and a total of 108 experimental units.
The experimental units were 5-L white plastic tubs containing 4 L of aged tap water. On 9 May, we placed six tadpoles into each tub. As with the TTD experiment, we set aside ten individuals from each population at the outset of the experiment to quantify the average developmental stage and mass of the individuals used in the experiment from each population. These initial masses ranged from 62 to 179 mg with development stages ranging from approximately 25 to 28.
We aged all tap water at least 24 hrs to allow chlorine to off-gas. For the sublethal salt treatment, we added lab-grade NaCl in aged tap water to reach a chloride concentration of 165.5 ± 0.3 mg Cl-/L (mean ± SE). This value represents the mid-range of chloride values in the ponds from which we collected the wood frogs. The tap water in the no-salt treatment had a background chloride concentration of 25.9 ± 0.3 mg Cl-/L (mean ± SE).
We fed the tadpoles a ration of Tetra goldfish food (Spectrum Pet Brands LLC) mixed with water ad libitum (starting at approximately 16.5 mg per tadpole per day and increasing up to approximately 22.2 mg per tadpole per day as the tadpoles grew) over the course of the experiment. We conducted water changes approximately every 4 d over the course of the experiment whenever the water appeared murky.
Nine days into the experiment, we quantified tadpole activity using scan sampling. Two observers made 10 observations each of every tub over the course of 75 min, which produced 20 observations of each tub. The observers slowly walked by the tanks and peered in so as not to disturb the tadpoles. For each tub, we recorded the number of individuals moving at a given moment. We conducted the observations prior to feeding, given that satiated tadpoles substantially reduce their activity to low levels. We used the mean percentage of individuals moving in each tub as our response variable. For the two tubs that had a tadpole not survive, we adjusted the percent activity to reflect the reduced number of tadpoles in that tub. Survival across the entire experiment was 99.7%.
The experiment was terminated after 22 d (31 May), when the tadpoles approached metamorphosis (i.e., Gosner stage 39). We then determined the mass and Gosner stage of each tadpole and used the means values of each experimental unit as our response variables.
